W3C home > Mailing lists > Public > w3c-sgml-wg@w3.org > May 1997

Re: Conformance + an apology

From: Tim Bray <tbray@textuality.com>
Date: Tue, 27 May 1997 06:40:28 -0700
Message-Id: <3.0.32.19970527063511.00a1fc40@pop.intergate.bc.ca>
To: w3c-sgml-wg@w3.org
At 08:49 AM 5/27/97 +0100, Digitome Ltd. wrote:
>With regard to that posting can anyone confirm or deny that
>a RAST/Grove/extended ESIS description of a parsed XML doc
>is desrable for parser dev. / conformance testing?

Early on in the XML process, we explicitly decided *not* to 
include an API in the XML spec.  The reason being that there is
no such thing as the universally correct API to a parser; that
depends on what the parser does.  We do try to be very precise
over what an XML processor has to pass an application - the
spec at the moment needs some improvement in this area.

>NXP and Lark parse XML. Do they produce the same result? How is
>this known?

We think they produce the same parse tree.  They do not have the
same API.  Notwithstanding the above comments, the various people
who have produced XML parsers, and in particular the people who
are starting to try to use them, are in strong agreement that
the APIs should be unified to the extent feasible.  I.e., if there
are N java-based XML parsers, and they all have an "Element"
class, then it would be nice if all the Element classes shared
some common constructors and methods.

Opinion: ESIS is useless.  Full groves/properties are too difficult
to fit into the spirit of XML.
Fact: At the moment, the correct place to have these discussions is
over in XML-DEV.

>In the months and years ahead, many products will claim XML
>parsing capabilities. How will such claims be gauged?

Fair question.  One point is that XML is simple enough that the
areas of ambiguity should be very small indeed. -Tim
Received on Tuesday, 27 May 1997 09:42:14 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:26 UTC